School of Mechanical Aerospace and Civil Engineering

3rd Year Fluid Mechanics

Inviscid Flows
T. J. Craft
George Begg Building, C41
Contents:
Navier-Stokes equations
Inviscid ows
Boundary layers
Transition, Reynolds averaging
Mixing-length models of turbulence
Turbulent kinetic energy equation
One- and Two-equation models
Flow management
Reading:
F.M. White, Fluid Mechanics
J. Mathieu, J. Scott, An Introduction to Turbulent Flow
P.A. Libby, Introduction to Turbulence
P. Bernard, J. Wallace, Turbulent Flow: Analysis Mea-
surement & Prediction
S.B. Pope, Turbulent Flows
D. Wilcox, Turbulence Modelling for CFD
Notes: http://cfd.mace.manchester.ac.uk/tmcfd
- People - T. Craft - Online Teaching Material
Introduction
One class of ows for which a simplied form of the Navier Stokes
equation can be considered is where the viscosity can be neglected.
These are known as inviscid ows.
Such ows might arise where the Reynolds number (Re =UL/) is high.
Since the Reynolds number represents a ratio of convective to diffusive
inuences, the limit of very high Re implies negligible diffusion.
In high-speed external ows around streamlined bodies, for example, the
ow well away from the body might be treated as inviscid (although the
approximation is not valid in the boundary layer or wake regions).
Boundary layer
Wake
Inviscid Flows 2010/11 2 / 22
In some cases solutions can be built
up by matching the inviscid solution
(in the outer ow region) to boundary
layer approximation solutions (see
later lectures) close to walls.
Boundary layer
Wake
At the other end of the scale from inviscid ows are creeping ows
(which will not be considered in this course).
These occur where the Reynolds number is very small (Re <<1), and
convection terms can thus be neglected.
U
t
+
U
2
x
+
UV
y
. .
=
1

P
x
+
_

2
U
x
2
+

2
U
y
2
_
(1)
Inviscid Flows 2010/11 3 / 22
The Euler Equations
If viscous effects are neglected, the governing equations are known as
the Euler equations.
In 2-D, incompressible, ow these can be written as
Continuity:
U
x
+
V
y
=0 (2)
Momentum:
U
t
+
U
2
x
+
UV
y
=
1

P
x
(3)
V
t
+
UV
x
+
V
2
y
=
1

P
y
(4)
or (making use of continuity)
U
t
+U
U
x
+V
U
y
=
1

P
x
(5)
V
t
+U
V
x
+V
V
y
=
1

P
y
(6)
Inviscid Flows 2010/11 4 / 22
The Stream Function
In 2-D we can dene a stream function, , such that the velocity
components are given by
U =

y
V =

x
(7)
Note that this denition ensures continuity is satised.
From the expressions in equation (7), the stream function can be
evaluated by integrating along a path in the uid:
=
_
B
A
Udy
_
B
A
V dx (8)
From equation (7),
U. =

y

y
=0
The gradient of is thus normal to the velocity vector, so does not
change in the direction following the uid velocity.
Inviscid Flows 2010/11 5 / 22
Hence is constant along a streamline.
Contours of therefore give the streamlines
of a ow.
=Const
=Const
The integral
_
B
A
Udy
_
B
A
V dx (9)
gives the total volume ow rate across a line between points A and B.
The difference in stream function values
between two streamlines therefore gives the
total volume ow rate between the lines.
=
=
1
2
Q=
1

2
Note that the above denition and properties of are not restricted to
inviscid ows.
Inviscid Flows 2010/11 6 / 22
In steady inviscid ow, subtracting /x of equation (4) (V momentum)
from /y of equation (3) (U momentum) leads to a partial differential
equation for :

x
(
2
) +

x

y
(
2
) =0 (10)
where
2
is the Laplacian operator
2
/x
2
+
2
/y
2
.
In principle this can be solved (with appropriate boundary conditions),
although numerical solution methods are usually required.
Having obtained , the velocity components can then be calculated from
equation (7).
Inviscid Flows 2010/11 7 / 22
Vorticity
In a 2-D ow the vorticity,
z
, is given by

z
=
U
y

V
x
(11)
This is associated with rotation of a uid element.
In inviscid ow there are no viscous shear stresses to deform uid
elements. It might, therefore, seem a reasonable approximation to
assume that the ow will be irrotational, ie.
z
=0.
This is an approximation that
can often be made in free
stream regions, away from
boundary layers.
Boundary layer
Wake
Inviscid Flows 2010/11 8 / 22
One can show that in an incompressible, inviscid, ow the vorticity is
constant along a streamline.
Taking /y of the U momentum equation (5) gives

t
_
U
y
_
+U

x
_
U
y
_
+
U
y
U
x
+V

y
_
U
y
_
+
V
y
U
y
=
1

2
P
xy
(12)
Since U/x +V/y =0 by continuity, we obtain

t
_
U
y
_
+U

x
_
U
y
_
+V

y
_
U
y
_
=
1

2
P
xy
(13)
Similarly, taking /x of the V momentum equation results in

t
_
V
x
_
+U

x
_
V
x
_
+V

y
_
V
x
_
=
1

2
P
xy
(14)
Subtracting equation (14) from equation (13) one obtains

z
t
+U

z
x
+V

z
y
=0 or
D
z
Dt
=0 (15)
Inviscid Flows 2010/11 9 / 22
Recall that the total derivative D/Dt represents the total rate of change as
a result of being convected with the ow.
Hence, if such a ow starts off irrotational (eg. at some far upstream
location where the ow is uniform, so
z
=0) it must remain irrotational.
From the denition of ,

z
=
U
y

V
x
=

y
_

y
_
+

x
_

x
_
=

2

x
2
+

2

y
2
(16)
The condition
z
=0 thus implies that the stream function satises the
Laplace equation,
2
=0.
Inviscid Flows 2010/11 10 / 22
Velocity Potential Function
For analyzing irrotational, inviscid, ow the velocity potential function, is
often used.
This is dened so that
U =

x
V =

y
_
and, in 3-D, W =

z
_
(17)
Note that this ensures
z
= U/y V/x =0.
Continuity then dictates that satises the Laplace equation:

x
2
+

2

y
2
=0 (18)
With appropriate boundary conditions, the above equation can be solved
to obtain the oweld for a particular problem.
Inviscid Flows 2010/11 11 / 22
Note the correct wall boundary condition for an inviscid ow is not simply
all velocity components being zero.
The wall-normal velocity component should be zero at a wall, but since
there is no viscosity the wall-tangential velocity need not be its value
arises as part of the problem solution.
From the (steady) U momentum equation, noting that V/x = U/y in
irrotational ow, we get

P
x
=U
U
x
+V
U
y
=U
U
x
+V
V
x
=

x
_
0.5(U
2
+V
2
)
_
(19)
Similarly, from the V momentum equation one gets

P
y
=

y
_
0.5(U
2
+V
2
)
_
(20)
Integrating the above equations we obtain
P +0.5(U
2
+V
2
) =Constant (21)
Inviscid Flows 2010/11 12 / 22
This is a special case of Bernoullis equation. The right hand side of
equation (21) is now constant throughout the ow, not simply along a
streamline.
Having obtained (and hence the velocities) from solving equation (18),
the pressure eld can then be determined from equation (21).
An attraction of the above approach is that the Laplace equation is
relatively easy to solve. There are some known analytical solutions, and
well-tested and inexpensive methods for numerical solutions.
The use of the velocity potential for irrotational ows can easily be
extended to 3-D ows, whereas the stream function cannot.
Inviscid Flows 2010/11 13 / 22
Example 1: Plane Channel Flow
Consider a channel ow developing between two parallel plates.
Appropriate boundary conditions are:

V =0, so /y =0, at y =0 and 1.

U =U
o
, so /x =U
o
, at x =0.

U/x =0, so
2
/x
2
=0, as x .
y=1
y=0
x=0
U
o
The function =U
o
x satises the Laplace equation, and the above
boundary conditions.
Differentiating gives the
velocity eld as U =U
o
and
V =0.
Potential contour lines and
streamlines are as shown.
0
.1
0
.2
0
.2
0
.3
0
.4
0
.5
0
.5
0
.6
0
.6
0
.7
0
.8
0
.9
0
.9
1
1
.1
1
.2
1
.2
1
.3
1
.4
1
.5
1
.6
1
.7
1
.8
1
.8
1
.9
X
Y
0 0.5 1 1.5 2
0
0.2
0.4
0.6
0.8
1
Without viscosity there is no boundary layer development.
Inviscid Flows 2010/11 14 / 22
Note that the streamlines and potential function contour lines are
perpendicular to each other.
This can be shown to be true generally:
. =

x

x
+

y

y
=VU+UV
=0
The gradient of is thus perpendicular to the gradient of , and it follows
that contours of must be perpendicular to contours of (streamlines).
Inviscid Flows 2010/11 15 / 22
Example 2: Plane Stagnation Flow
In this case it is convenient to work in polar coordinates (r , ).
The Laplace equation becomes
1
r

r
_
r

r
_
+
1
r
2

2
=0 (22)
x=0
y=0
There are a family of solutions to this of the form
=Ar
n
cos(n) for constants A and n (23)
For simplicity, take A =1.
n =1 gives uniform ow.
n =2 gives a ow
impinging onto a at plate,
with potential lines and
streamlines as shown.
-0.8
-0.7 -0.7 -0.6
-0.6
-0.5
-0.4
-0.4
-0.3
-0.3
-0.3
-0.2
-0.2
-0.2
-0.1
-0.1
-0.1
0
0
0
0
0
.1
0.1
0.1
0
.2
0.2
0
.2
0
.2
0
.3
0
.3
0
.4
0
.4
0
.5
0
.5
0
.6
0
.6
0
.7
0
.7
0
.8
0
.9
X
Y
-1 -0.5 0 0.5 1
0
0.2
0.4
0.6
0.8
1
Inviscid Flows 2010/11 16 / 22
In cylindrical polar coordinates the velocity components are given by
V

=
1
r

and V
r
=

r
(24)
The radial velocity, V
r
, in this ow is thus

r
=2r cos(2) (25)
At =90
o
, this gives /r =2r

So the velocity along the stagnation line decreases linearly towards

the wall.
Along the impingement wall ( =0
o
) we get /r =2r

So the velocity along the wall increases linearly with distance from
the stagnation point.
Flow deections through angles other than 90
o
can be obtained by taking
other values of n.
Inviscid Flows 2010/11 17 / 22
Example 3: Flow Past a Cylinder
Consider a uniform stream of velocity U
o
, owing past a circular cylinder
of radius R as shown.
Working in cylindrical coordinates, potential
ow in such a case can be represented by
=U
o
(r +R
2
/r )cos() (26)
R
U
o
The radial and circumferential velocity components are given by
V
r
=

r
=U
o
(1R
2
/r
2
)cos()
V

=
1
r

=U
o
(1+R
2
/r
2
)sin()
In the far eld (large r ), this gives
V
r
=U
o
cos() and V

=U
o
sin()
corresponding to uniform ow in the x direction.
Inviscid Flows 2010/11 18 / 22
On the cylinder r =R, we get
V
r
=0 and V

=2U
o
sin()
The potential function contours and streamlines are as shown.
X/R
Y
/
R
-4 -2 0 2 4
-2
-1
0
1
2
There is no ow separation, and the streamline patterns are perfectly
symmetric fore and aft of the cylinder.
Inviscid Flows 2010/11 19 / 22
The cylinder surface pressure can be obtained from equation (21):
P +0.5V
2

=Const =0.5U
2
o
(27)
Thus
P =0.5U
2
o
2U
2
o
sin
2
()
=0.5U
2
o
(14sin
2
())
The pressure is perfectly symmetric
fore and aft, and the cylinder
experiences no drag force! 0 30 60 90 120 150 180
Theta
-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
C
p
This example illustrates a limitation of the inviscid, irrotational,
approximations.
In reality they do not hold near the cylinder surface, where boundary
layers form, leading to ow separation under adverse pressure gradient
conditions.
Inviscid Flows 2010/11 20 / 22
Superposition of Solutions
Since the Laplace equation is linear,
solutions to it can be added together, or
superimposed.
For example, the potential function
=k, for constant k, gives a vortex
centered at the origin.
X
Y
-2 0 2
-2
0
2
This could be added to the
function used in example 3,
giving ow around a
cylinder with circulation:
=U
o
(r +R
2
/r )cos()+k
X/R
Y
/
R
-4 -2 0 2 4
-2
0
2
Inviscid Flows 2010/11 21 / 22
Summary
In 2-D ows the stream function can be dened to aid in ow analysis.
Inviscid ows are governed by the Euler equations.
In an inviscid, irrotational, ow the velocity potential function can be
dened.
In potential ow problems the streamlines and potential function contours
are orthogonal to each other.
Inviscid ow approximations can typically be applied to high speed ow
away from solid surfaces. Such solutions can, in some instances, be
matched to approximate boundary layer solutions for the near-wall region.
Inviscid Flows 2010/11 22 / 22